Trigger Happy

Shoot first, ask questions later

- STEVEN POOLE

For many years, a small cult of 'existential risk' thinkers has been worrying itself half to death in San Francisco about the possibility (or, as they see it, the very high probability) that a future artificial superintelligence will destroy all humans. I guess it's nice that they have a hobby, but have you seen ChatGPT 5? Shouldn't we be more worried about the actually-existing robotic dogs?

There is something genuinely unnerving about what has become the default design of the canine robot: headless, with spindly legs, but extremely agile and strong. The general look was laid down by Boston Dynamics, which calls its robot dog Spot in a blatant attempt to make it seem less terrifying. If you acquire a taste for robot-dog videos, as I did recently, you can watch similar machines sprinting through forests and up hills of scree, being whacked with a large stick by a human and barely wobbling, emerging threateningly from a river, or crossing a street while balancing on an exercise ball.